The advisor's target allocation is 40% domestic equities, 35% bonds (10% of the 35% is foreign), 15% foreign stocks and 10% Alternatives which includes managed futures. Anyone may or may not agree with the allocation but like any allocation this will be fine most of the time but it may make sense to raise cash at different points in the future.
The way the advisor implements the model is problematic. As you can see all of the holdings are actively managed mutual funds. The funds chosen are probably "good" funds. Advisors don't usually pick the worst performers in the group. "The small cap fund I chose for you has lagged its benchmark index and 80% of all small caps for the last ten years," doesn't happen too often.
Boilerplate always warns about past performance but what else can you do? Any buyers of actively managed funds out there go with the active fund that is 15 basis points cheaper or do the take the better track record?Additionally there is no way to do any sort of forward looking analysis. What will the manager of your active mutual fund want to own one year from now? As the question is unanswerable forward looking analysis is undoable. The potential consequence is that the portfolio owns a bunch of funds where the managers all draw similar conclusions and so invest their funds similarly. This is fine if they are right or a potential deathblow if they are catastrophically incorrect like being overweight tech in 2000 or overweight financials in 2008.
This problem doesn't really have the same consequence with narrow based funds. If a portfolio allocates 15% to an actively managed financial fund and fills out the rest of the portfolio with sector funds then the portfolio isn't going to get caught with 30% in financials. The problem of too much of one sector or not enough of another can occur using actively managed country funds.
Using an actively managed fund as part of a diversified portfolio can certainly make sense but total reliance seems like a problem waiting to happen.
33 comments:
Roger or others, are there any world ETFs done right? I'm looking for something simple for a family member where they can get domestic and foreign exposure in one ETF. I'd consider two funds max, but country specific exposure isn't really an option. And when I say "world" I'm probably looking at places that aren't stagnant or highly correlated to the US (which probably rules out Europe and Japan). Any ideas here?
Thanks.
Roger: The "Manager's Portfolio" from the WSJ article contains 18 funds. All this over-complexity accomplishes is to impress the client with the manager's "expertise." I would fire the manager and invest as follows:
40%-Vanguard Total Stock Index (VTSMX)
25%-Vanguard Total Bond Index (VBMFX)
10%- T. Rowe Price International Bonds(RPIBX)
20%-Vanguard FTSE Ex-US (VFWIX)
5% - Vanguard Emerging Markets Index (VEIEX)
Five low cost funds for 18 funds with low to high expenses, no worries about fund allocation decisions. I guess this merely confirms what you are saying, Roger, about the potential dangers of a portfolio of all actively managed funds. In the WSJ portolio, there are really two layers of manager activity - the fund manager's and the investment manager - management squared, if you will.
Bill B,
Are you looking fora world fund that includes the US and Europe (and everywhere else for that matter) or are you looking for a foreign growth or emerging market fund?
I like VWO which is an emerging market fund from vanguard, but would be considered highly speculative and inappropriate as a single fund or one of two funds in a portfolio. Still I like emerging markets going forward and believe it has a place in every bodies portfolio (probably excessively weighted in my portfolio according to some people).
On the issue of correlation, I think we are in a period of very high correlation among most assets.
BillB I've never used an all-world fund and addressing the rest of the question can be construed as a recommendation which becomes a compliance issue, apologies.
JimL if that works for you, great.
Ya, I probably wasn't real clear. I was looking for world exposure (including U.S.). I accomplish this myself with SPY/PBP (for US) and VWO for foreign. I may just go w/ this approach, but thought I might be able to simplify it just a bit more for someone who wants to set and forget.
Thanks for the suggestions so far.
Good post Roger...I agree with Jim that the portfolio is amazingly complex that only intends to impress rather than create return. However, I disagree with his alternative portfolio of Vanguard funds. With the "Modern Portfolio Theory" fraud finally being exposed, a tactical approach utilizing ETFs is a much more nimble approach not only allowing instant sells (&/or buys)in a fast market, many have options writing available to add to the mix.
tactical is not right for all. Me yes and you too but not everyone. Modern portfolio theory still works. Well, except for when it doesn't. Oops
Bill B,
There is always ACWI from iShares and VT from Vanguard for all-world exposure (both about 45% US, 40% developed, and 15% emerging).
If you then add in some PBP you both reduce the foreign exposure (55% is considered large by some) and add in some volatility reduction.
I have heard a lot of people talk about developed large-cap stocks of all domiciles becoming tightly correlated, globalization being the reason given. If you consider that most large global companies do business all over the globe, then this seems reasonable and plausibly likely to continue.
Scott
Jim L and Paul,
Putting aside the question of "tactical" trading vs "set and forget", why are you both against portfolios with large numbers of funds? Do you have some underlying rejection of the utility of rebalancing?
If rebalancing works, then what reason could there be not to split up (for example) my small-cap exposure by region, and my bond exposure by bond type and duration? Besides small increases in fees?
Scott
I much prefer etfs to mutual funds, but one day I bought VEIEX from Vanguard instead of VWO which I own a lot more of. These are suppose to be the same funds so to speak with VWO being and ETF version of the fund.
VEIEX has a 2% redemption penalty if sold within 6 months, which is one reason I prefer VWO. But the day I wanted to allocate money to the markets I thought most of the ETFs were over priced to net assets a tad. So on a high volume buying day I thought the mutual fund was a better buy.
Just a nuance that should be remembered in my opinion from time to time even though I am a bigger fan of etfs in general
Not against balance Scott, just have no use for funds in a professionally managed portfolio.
And Roger - MPT has never been reliable, only lucky! Even Markowitz later in life questioned the soundness of his theory.
Scott: I'm not against large numbers of funds in principle, but I think that "the fewer the better" rule applies. In managing my own portfolio around core index funds, I have found it hard to resist adding specialty funds, like emerging small cap growth, microcap, and natural resource funds. My basic problem with "over-complexity" arises when it is recommended by so-called experts who I believe often enhance their livelihood from increased complexity without benefit to the investor.
MPT was very good science for the 1950s because the computational requirements were fairly low. However it is severely flawed and due to be replaced:
Flaw #1: Average period returns for assets are used instead of compound returns. (For high volatility assets the average (mean) return is substantially higher in value than the compound return. This means that MPT based portfolio optimizers called “Mean-Variance Optimizers” put too much weight into volatile assets.)
Flaw #2: MPT equates standard deviation of returns with “risk”. Since this is a false relationship, the resulting portfolios should be expected to be inferior.
Flaw #3: MPT thinks that correlations during good times are important – but they aren’t at all! Nobody cares if assets are un-correlated during good times because this doesn’t keep everything from plunging during a crash.
Since these flaws largely apply to post-modern portfolio theory as well what is the solution? Simple: just optimize for compound returns and draw-downs instead of mean returns and standard deviation.
The guru's approximate portfolio: (40.0% MKT-TSM , 35.0% Bonds , 10.0% COMM , 5.0% Pacific , 5.0% Intl Value , 5.0% EM) has compounded at 10.2% since 1972 with the worst 2 years being 2008 @ -24% and 1974 @ -11.9%.
If you re-weight the portfolio according to the optimization framework I suggested above then you end up with: 73.7% ITB , 15.0% EM , 8.1% Pacific , 3.3% Intl Value (note that I capped emerging at 15% exposure)
This portfolio has also compounded at 10.2% since 1972. The two worst years were 2008 @ -8.6% and 1973 @ -2.2%.
MPT and guru ad-hoc portfolios can definitely be improved upon! However most individuals (not readers of this blog I'm sure) hold portfolios that are probably far inferior to an MPT or standard guru allocation ;)
What "guru" would this be?
Roger - it's common for magazines like Money and Kiplinger's Personal Finance to do this for people or with people's existing portfolios. They use managed funds to create total portfolios. It sure does bring in a lot of variables.
@Stephen, by "guru" I meant Ms. Debra Brede who designed the portfolio from the WSJ article that Roger mentioned.
You are right that many of the magazine gurus seem to have a similar approach: stick with around 40% bonds since they remember hearing something about MPT one time. Then pack the rest of the portfolio with active funds just intriguing enough to get magazine inches, but not daring enough to perform any differently than the policy portfolios of other gurus ;>
The sub-title on the WSJ piece mentions "managed futures" which do make an appearance at a 3% weighting! But the editorial staff could have also gone with the subhead of "40% bond portfolio with high expenses still outperforms what our average reader holds". ;->
MPT was very good science for the 1950s because the computational requirements were fairly low. However it is severely flawed and due to be replaced
I don't mean to sound snarky, but "due to be replaced" is a bit presumptuous. Also, when and by what? And how does more computing power have anything to do with it? Computing power has increased a ridiculous amount since the 50's and there hasn't been much (any?) progression on MPT.
What are some examples of portfolios that are superior to MPT? I'm not saying MPT is or isn't lucky (it's had a hell of a "lucky" run so far), I would just enjoy thinking about other ways.
FWIW, I build trading systems that are somewhat complicated. I think in the short term, I can have a good gauge of probability. Once I extend beyond short term, it starts to perform closer to the market. If one could combine the probabilities of short term into long term holdings, I'd be all ears.
Not disagreeing because I'm an MPT die hard, just a bit curious and skeptical :)
Matthew,
The first two flaws you point out with MPT (use of mean and stddev) are oft-discussed, and your proposed remedies seem quite reasonable. However, I always thought that what MPT was really *about* was the framework of optimizing risk/return profiles, and not so much the metrics used.
As for flaw #3 (correlation spikes during crashes), I don't see how changing your metrics helps much :). Any asset allocation involving risky assets is going to suffer losses when people suddenly want to avoid all risk. I suppose this explains why all the "optimal-risk" portfolios I've seen -- including yours -- have around 70% mid-term treasuries.
In any case, would you be willing to share exactly how you calculate your "drawdown" metric? I would be quite interested.
Scott
Hi Bill B, I am sure I am over my quota for comment length today - but I will try to answer some of your questions.
I think that MPT should be replaced in practice by a concept I call RiskCog. (I am not sure that I should presume to name it since it is so simple it must have been independently discovered by others.)
RiskCog uses geometric returns and portfolio losses where MPT uses arithmetic returns and variance (of arithmetic returns) as proxies for the concepts "return" and "risk".
Optimizing over this alternate framework requires more computational resources than MPT and would have been only theoretically possible in the 1950s.
For comparison here is an example MVO portfolio taken from a publication by a popular financial software vendor and compared to RiskCog using data from 1972 to 2006:
*Mean-Variance Optimal Portfolio
Total US Stock Market 48.1%
Foreign Stocks 32.2%
US Bonds 14.9%
Commodities 3.5%
REITs 1.3%
T-Bills 0.0%
CAGR '71-'06 11.4%
CAGR '71-'08 9.7%
2008 loss -32.7%
*RiskCog Optimal Portfolio
US Bonds 39.6%
Commodities 27.7%
REITs 19.1%
Foreign Stocks 6.8%
Total US Stock Market 6.7%
T-Bills 0.0%
CAGR '71-'06 11.4%
CAGR '71-'08 10.1%
2008 loss -22.5%
These are buy-n-hold portfolios created based on historical data so the normal limitations and warnings for that portfolio class apply. I am currently working on an optimization approach that will allow an investor to decide how much to allocate to different active trading systems.
I wouldn't recommend this particular RiskCog optimal portfolio actually since it doesn't contain precious metals, long bonds, or emerging markets. In fact I would prefer to hold 1/3 each of those excluded asset classes than the portfolios above - but I couldn't get that published in the WSJ ;->
apologies for verbosity today!
Interesting. No domestic stocks on that portfolio?
The only way you can backtest for an actual investor is to use Vanguard mutual funds, all of which go back to 1994 for those asset classes.
Right off the bat I'm worried 'cause my 2008 calculation only shows a loss of 2%, which is significantly off from yours.
THis is kind of like a "Larry Swedroe" portfolio discussion we'have had here a few times. Roger will occasionally refer to, I think, Taleb. The "put the big majority of the portfolio in Treasuries and the rest in whatever you like that's risky" hence, I assume you're choices that are all int'l.
The portfolios results are mostly similar, since the 70% (or more) are the same - treasury bonds. I'd guess yours is a tad higher since you are choosing intermediate term treasuries and not short term treasuries.
@Matthew . Fantastic! Thank you!
OK, found a riskcog site. You use this, Matthew? Anyway, something interesting here, if you include gold, your drawdown shrinks pretty nicely no matter what else is in your portfolio. Many have noted that gold tends to zig when everything else is zagging. Maybe the next time everything else zigs, gold doesn't zag. This also seems to be the linchpin of Permanent Portfolio advocates as well. I tend to wonder if gold is the next thing to "fail" the next time things go south.
I've probably reached my ramble limit for today as well, but this has been very thought provoking stuff.
I have been saying we are turning Japanese for over 2 years here. Now Hussman is singing the song so I guess everyone will agree with me now.
What do you think the chances are that "Matthew" is the owner of "riskcog.com"?
Still, nice looking tool.
Scott,
Matthew has been contributing in a meaningful way to the conversation for months. In that light it is not a breach of etiquette to promote a site with that sort of contribution behind it.
BTW, no idea whether it is his site or not.
Matthew, I apologize -- Roger is right. I enjoy this site but don't read/post comments very often, so I have no right to make or imply judgments. I did enjoy your posts today.
Yeah, I built that site. Thanks for the encouragement guys.
@Matthew, GREAT! Sign me up as a tester then :) Oh, and nice work :)
@SD, you're right, this does start to lend some validation to the doom and gloomer types. If I just do 5-yr treasuries with emerging markets, I can really see my risk being controlled with only two inputs. Simple = good!
I'm still very nervous about gold. It's like the magic powder that automatically makes any portfolio better (less drawdown). Surely that can't ALWAYS be the case forever.
@Matthew -- I apologize again, and after playing with the site a bit more, it really is a very nice piece of work, and generous of you to share it. I hope the word spreads.
Billb - that's exactly what I end up thinking about that type of portfolio. Ok, you go short term treasuy or int. term treasury for the 70%, then you can think of all kinds of combinations for the other 30%. I think my calculations show that 70% treasury/30% EM actually does a little better than Matthew's portfolio. I'd post results but I haven't figured out how to easily cut/paste from Excel into Blogger.
What I probably need to do is start doing these calcs vertically on the spreadsheet rather than horizontally.
@Stephen if you want it is possible to link to portfolio allocations from the site I created by copying the url. Here is one with 70% treasuries.
http://www.riskcog.com/portfolio.jsp#5jwf8ji
Allocation: 70% TBILL, 30% EM
Compound return = 9.84%
Worst year: 2008 -14.76%
I am not sure which of the portfolios your are referring to by as "Matthew's Portfolio". Here is probably the best one I mentioned in passing today:
http://www.riskcog.com/portfolio.jsp#59h89hi9hc
Allocation: 33.3% GOLD, 33.3% EM, 33.3% LTGB
Compound return = 13.19%
Worst year: 1981 -9.21%
Allocation does not matter, if you are investing with an active fund managger then you will lose much of your gains through fees - fees upon fees upon feees upon fees.
All you need to do is open a Vanguard account and put your money into a target fund such as 2020, 2030, 2040, 2050 and stop getting sucked into giving your money to active managers / sales people for wall-street.
Post a Comment